Navam Logo
Navam
ACTIVE PROJECT - SPONSOR DEVELOPMENT

Sentinel Visual-First Agent Testing

Make AI agent testing as intuitive as Postman made API testing. Fork it commercially, sponsor active development.

v0.22.0 473 tests 68 components BIS License
View Code

The Problem with Testing AI Agents

Testing AI agents today requires custom Python code, manual validation, or writing complex test scripts. Non-developers can't participate. Collaboration is difficult.

Manual Testing

Slow & Error-Prone
  • ✗ Copy-paste prompts into ChatGPT
  • ✗ Manually check each response
  • ✗ No version control or collaboration
  • ✗ Hard to detect regressions

Code-Only DSL

Requires Expertise
  • ✗ PMs can't write tests
  • ✗ QA needs to learn YAML syntax
  • ✗ Hard to visualize test structure
  • ✗ Steep learning curve
SENTINEL APPROACH

Sentinel Visual

Point, Click, Test
  • ✓ Anyone can build tests visually
  • ✓ Real-time YAML generation
  • ✓ Git-friendly for collaboration
  • ✓ Production-ready from day one

The Sentinel Workflow

Build tests by clicking components, connecting nodes, and seeing YAML generated in real-time. No coding required.

PRODUCTION-READY CANVAS
Sentinel Visual Canvas with Component Palette and YAML Preview
1

Click Components

Click Input, Model, or Assertion nodes to add them to the canvas

2

Connect Nodes

Drag from node handles to create test flow connections

3

See YAML

Watch type-safe YAML generate in real-time as you build

4

Download & Run

Copy or download YAML for version control and execution

Active Development Roadmap

We're building in public. Sponsor to vote on priorities and shape the future of Sentinel.

SHIPPED

v0.22.0 - Unified Library Tab

Released November 23, 2025

  • ✓ Unified Library Tab with templates and user tests
  • ✓ 12-category system (Q&A, Code Gen, Browser Agents, Multi-turn, etc.)
  • ✓ 473 frontend tests passing with comprehensive coverage
  • ✓ 68 React components across multiple categories
IN PROGRESS

v0.23.0 - Model Execution & Result Storage

Target: Q1 2026

  • → Live execution dashboard with real-time progress
  • → Result storage and test run history (SQLite)
  • → Metrics collection and analytics
  • → Performance tracking and comparison
PLANNED

v0.24.0 - Additional Model Providers

Target: Q1 2026

  • ○ Amazon Bedrock integration (multi-model support)
  • ○ HuggingFace API support (hosted endpoints)
  • ○ Ollama integration (local models)
  • ○ Provider comparison and cost tracking

Beyond v0.24.0

v0.25.0 - LangGraph Framework Support

State machine testing and multi-agent workflows

v0.26.0 - AI-Assisted Test Generation

Generate tests from natural language descriptions

v0.27.0 - Visual Assertion Builder

Enhanced visual forms for creating assertions

v0.28.0 - Regression Detection

Comparison view and trend analysis

v0.29.0 - Collaborative Workspaces

Team testing and shared test suites

v0.32.0 - CI/CD Integration

Automated testing and deployment pipelines

Sponsors can vote on priorities! Help us decide what to build next.

Production-Ready Features

5 Node Types

Input, System, Model, Tool, and Assertion nodes for comprehensive test coverage

Real-time YAML

Watch type-safe YAML generate as you build. Copy or download for version control

473 Tests Passing

Production-ready with comprehensive frontend test suite covering all components and features

React Flow 12.3

Production-ready canvas library with 19K+ stars. Battle-tested by Stripe, AWS

Tauri 2.0 Desktop

Native desktop app for Windows, macOS, Linux. Lightweight and fast

Git-Friendly

Clean YAML diffs for collaboration. Visual changes = readable version control

Get Started in 5 Minutes

Installation

1. Clone Repository

git clone https://github.com/navam-io/sentinel.git
cd sentinel/frontend

2. Install Dependencies

npm install

3. Launch Desktop App

npm run tauri:dev

Requirements

  • Python 3.10 or higher
  • Node.js 18 or higher
  • 4GB RAM minimum

Frequently Asked Questions

What is Sentinel?

Sentinel is a visual-first testing platform for AI agents that makes testing as intuitive as Postman made API testing. Build tests by clicking components on a canvas or write declarative YAML specs—your choice.

Why sponsor Sentinel?

Sponsoring supports active development while giving you exclusive benefits: private AI engineering newsletter, ability to influence roadmap priorities, early access to new features, and direct support channel. Plus you can fork and use commercially (BIS license) regardless of sponsorship.

Can I use Sentinel commercially without sponsoring?

Yes! Sentinel is BIS licensed. You can fork it, modify it, and use it commercially without any restrictions or fees. Sponsorship is optional and provides additional benefits for those who want to support active development.

What's the difference between Sentinel and writing tests in code?

Sentinel provides a visual canvas where you click to add components, see the test structure visually, and get real-time YAML generation. This makes it accessible to non-developers (PMs, QA, researchers) while still providing the power of code (YAML DSL) for version control and collaboration.

What operating systems does Sentinel support?

Sentinel runs on Windows, macOS, and Linux as a Tauri 2.0 desktop application. It requires Python 3.10+ and Node.js 18+.

How is Sentinel different from other testing tools?

Sentinel is specifically designed for AI agents, not traditional software. It understands prompts, models, tools, and agent-specific assertions. The visual canvas makes it accessible to non-developers while the YAML DSL provides power for engineers.

Learn More About Sentinel

Sentinel Visual Canvas

Building Sentinel's Visual Canvas

How we shipped v0.3.1 with a production-ready visual canvas in 48 hours—migrating from Svelte to React, simplifying UX from drag-and-drop to click-to-add, and building comprehensive test infrastructure.

Read the full story

Start Testing AI Agents Visually

Fork Sentinel for free (BIS license) or sponsor to support active development and receive exclusive benefits. Your choice.

View Code on GitHub